The fans flare, the lights come up — and two seconds later, darkness. Or it lives for a minute and then cuts out like someone pulled the plug. A PC that starts and then stops is not dead; it is being switched off, by something that thinks it is protecting the machine, or by a power supply that cannot hold the load. The biggest clue is free: how long does it live?
Most no-start symptoms are vague, but this one carries a timestamp, and the timestamp separates the suspects. A death inside the first few seconds usually means a protection trip or a processor overheating almost instantly. A death after a minute or two is heat building up. A death only under load — a game opening, a heavy program — is a different story again, told elsewhere. This page stays at the first-check level: read the timing, check the cooling you can see, and know when the timing is telling you to stop restarting it.
A modern processor with its cooler loose, unlatched or missing its fan connection protects itself the only way it can: it shuts the machine down within seconds, before the heat can hurt it. So a two-second death after a move, a clean, or any internal work is the cooler until proven otherwise. Power off, unplug, and look: Is the cooler's fan cable on its header (labelled CPU FAN)? Do the cooler's four feet or screws sit evenly latched? Did the whole heatsink get knocked crooked? A cooler refit — sometimes with fresh thermal paste if the old contact broke — is the entire repair surprisingly often.
If the machine survives the first stretch and dies as it warms, picture the inside honestly: a mat of dust felted over the heatsink fins, a case fan that stopped years ago, a PC pushed into a closed cabinet. Feel the exhaust air, look through the vents with a torch, and check every fan you can see actually spins at start. Cleaning here means unplugged, gentle air bursts, fans held still — and dust mats lifted out, not blown deeper in. If the shutdown time grows longer after a clean, heat was indeed confessing.
Some supplies hold the easy idle loads and collapse when the processor and graphics card ask for real current — the machine then dies at the same demanding moment, like a game launching. At the first-check level your evidence is pattern and age: an old, budget, or recently-stressed supply, deaths that track load rather than time. The honest test remains the swap with a known-good supply; there is no software that can prove a hardware power fault, and no setting that fixes one. Keep this suspect, but check the cooler first — it is free.
Protection also trips on electrical shorts. The classic after a rebuild or clean: a loose screw that rolled under the motherboard, or a damaged USB device drawing the port down. Strip the test down — unplug every USB extra, printers, drives, dongles — and try again. A machine that now stays alive has introduced you to its culprit; plug extras back one at a time. If the cabinet was recently opened and you can hear anything loose rattling, find it before the next power-up, not after.
Do not power-cycle it twenty times hoping for a better roll of the dice. Each forced start into a real heat or short problem adds stress, and if the storage drive is part of the story, repeated hard cuts are unkind to it too. After three or four timed attempts, write down the timing and the pattern and stop. That note, plus which of these checks you ran, is the difference between a technician starting at zero and starting at your evidence. A start-then-stop machine usually has a findable cause — but it is found by timing and looking, not by persistence.
The video runs the first checks in order — and on start-up problems, the timing trick narrows the hunt before a single part is swapped.
▶️ Watch Full VideoConsistency is a clue: the same survival time means the same trip point. Within seconds, the usual story is the processor protecting itself — a loose or disconnected CPU cooler or its fan — or an electrical protection trip. Check the cooler's seating and fan cable first, then look for shorts and loose screws if the case was opened recently. Random timings point more at power supply or building heat instead.
A handful of timed attempts for evidence, yes. Twenty hopeful restarts, no. If the cause is heat, each attempt bakes the processor again before its protection trips; if it is a short, each attempt stresses the board. Time the death, note the pattern, run the checks on this page — and if it still dies, the note you wrote is worth more than another ten tries.
Not at this stage of the start. Shut-downs in the first seconds happen before Windows or any software has loaded — no virus gets a vote that early. Software can cause restarts and shutdowns after the system loads, which is a different symptom family with a different hunt. A machine dying in two seconds has a hardware story: protection, heat, or power.